我改进了尼尔斯的解决方案

class Array          
  def except(*values)
    self - values
  end    
end

现在你可以使用

[1, 2, 3, 4].except(3, 4) # return [1, 2]
[1, 2, 3, 4].except(4)    # return [1, 2, 3]

以下是一些基准:

require 'fruity'


class Array          
  def rodrigo_except(*values)
    self - values
  end    

  def niels_except value
    value = value.kind_of?(Array) ? value : [value]
    self - value
  end
end

ARY = [2,4,6,3,8]

compare do
  soziev  { a = ARY.dup; a.delete(3);               a }
  steve   { a = ARY.dup; a -= [3];                  a }
  barlop  { a = ARY.dup; a.delete_if{ |i| i == 3 }; a }
  rodrigo { a = ARY.dup; a.rodrigo_except(3);         }
  niels   { a = ARY.dup; a.niels_except(3);           }
end

# >> Running each test 4096 times. Test will take about 2 seconds.
# >> soziev is similar to barlop
# >> barlop is faster than steve by 2x ± 1.0
# >> steve is faster than rodrigo by 4x ± 1.0
# >> rodrigo is similar to niels

再次使用包含大量重复项的更大数组:

class Array          
  def rodrigo_except(*values)
    self - values
  end    

  def niels_except value
    value = value.kind_of?(Array) ? value : [value]
    self - value
  end
end

ARY = [2,4,6,3,8] * 1000

compare do
  soziev  { a = ARY.dup; a.delete(3);               a }
  steve   { a = ARY.dup; a -= [3];                  a }
  barlop  { a = ARY.dup; a.delete_if{ |i| i == 3 }; a }
  rodrigo { a = ARY.dup; a.rodrigo_except(3);         }
  niels   { a = ARY.dup; a.niels_except(3);           }
end

# >> Running each test 16 times. Test will take about 1 second.
# >> steve is faster than soziev by 30.000000000000004% ± 10.0%
# >> soziev is faster than barlop by 50.0% ± 10.0%
# >> barlop is faster than rodrigo by 3x ± 0.1
# >> rodrigo is similar to niels

甚至更大，有更多的副本:

class Array          
  def rodrigo_except(*values)
    self - values
  end    

  def niels_except value
    value = value.kind_of?(Array) ? value : [value]
    self - value
  end
end

ARY = [2,4,6,3,8] * 100_000

compare do
  soziev  { a = ARY.dup; a.delete(3);               a }
  steve   { a = ARY.dup; a -= [3];                  a }
  barlop  { a = ARY.dup; a.delete_if{ |i| i == 3 }; a }
  rodrigo { a = ARY.dup; a.rodrigo_except(3);         }
  niels   { a = ARY.dup; a.niels_except(3);           }
end

# >> Running each test once. Test will take about 6 seconds.
# >> steve is similar to soziev
# >> soziev is faster than barlop by 2x ± 0.1
# >> barlop is faster than niels by 3x ± 1.0
# >> niels is similar to rodrigo

我喜欢在其他答案中提到的-=[4]方式来删除值为4的元素。

但有这样一种方法:

[2,4,6,3,8,6].delete_if { |i| i == 6 }
=> [2, 4, 3, 8]

在“基本数组操作”中提到的map函数之后。

在ruby中编译所有不同的delete选项

delete -按值删除匹配的元素。如果有多个值匹配，则删除所有值。如果您不关心出现的次数或确定单个出现，则使用此方法。

a = [2, 6, 3, 5, 3, 7]
a.delete(3)  # returns 3
puts a       # return [2, 6, 5, 7]

delete_at -删除给定索引处的元素。如果你知道索引，就用这个方法。

# continuing from the above example
a.delete_at(2) # returns 5
puts a         # returns [2, 6, 7]

delete_if -删除block为true的每个元素。这将修改数组。数组在调用块时立即改变。

b = [1, 2, 5, 4, 9, 10, 11]
b.delete_if {|n| n >= 10}.  # returns [1, 2, 5, 4, 9]

reject -该函数将返回包含给定块为false的元素的新数组。顺序是用这个来维持的。

c = [1, 2, 5, 4, 9, 10, 11]
c.reject {|n| n >= 10}.  # returns [1, 2, 5, 4, 9]

拒绝!-同delete_if。数组在调用块时可能不会立即改变。 如果你想从数组中删除多个值，最好的选择如下所示。

a = [2, 3, 7, 4, 6, 21, 13]
b = [7, 21]
a = a - b    # a - [2, 3, 4, 6, 13]

借用Travis的评论，这是一个更好的答案:

我个人喜欢[1,2,7,4,5]-[7]，结果=>[1,2,4,5]从irb

我修改了他的答案，因为3是他示例数组中的第三个元素。对于那些没有意识到3在数组中的位置2的人来说，这可能会导致一些困惑。

另一个选择:

a = [2,4,6,3,8]

a -= [3]

结果是

=> [2, 4, 6, 8]